[Ian Bell]

A while ago we were discussiig theoretical and actual measured noise in the
first stage of a preamp. RDH suggests the equivalent input noise resistance
for a triode is about 2.5/gm. Patrick noted he had never seen noise
anywhere near as low as this.

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

which suggests the 2.5/gm only applies at frequencies above 10KHz and that
below this 1/f noise dominates. It also suggests this can be reduced by
choosing very pure materials for the cathode nickel, and running the tube
at an optimized (generally low) anode current.

Don't know who the author is and as a rule I take anything on wikipedia with
a large grain of salt. Comments?

Cheers

ian

[Phil Allison]

"Ian Bell"

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

which suggests the 2.5/gm only applies at frequencies above 10KHz and that
below this 1/f noise dominates.


** Not what it says at all.

" For special low noise audio tubes, the frequency at which 1/f noise takes
over is reduced as far as possible, maybe to something like a kHz. "

Applies to valves like the 12AX7 and 12AY7 - their noise spectrum is
pretty much white noise, where the energy is *proportional* to bandwidth. A
gradual rise in noise below 1kHz is insignificant, since it is so small in
the first place.

BTW:

The section about " Class AB and B " is a load of ********.

The author say he got his info from " The Audio Amateur" and "
Stereophile" magazine.

Bout as reliable as the weather.....



......... Phil

[Eeyore]

Phil Allison wrote:

"Ian Bell"

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

which suggests the 2.5/gm only applies at frequencies above 10KHz and that
below this 1/f noise dominates.

** Not what it says at all.

" For special low noise audio tubes, the frequency at which 1/f noise takes
over is reduced as far as possible, maybe to something like a kHz. "

Applies to valves like the 12AX7 and 12AY7 - their noise spectrum is
pretty much white noise, where the energy is *proportional* to bandwidth. A
gradual rise in noise below 1kHz is insignificant, since it is so small in
the first place.

BTW:

The section about " Class AB and B " is a load of ********.

The author say he got his info from " The Audio Amateur" and "
Stereophile" magazine.

Bout as reliable as the weather.....

Hmmmm, the author reckon class AB and B amps produce 'aharmonic distortion' !
There's no explanation either of what the terms mean !

Talk about the blind leading the blind !

Graham

[Chris Hornbeck]

On Thu, 21 Jun 2007 23:06:13 +0100, Ian Bell
wrote:

A while ago we were discussiig theoretical and actual measured noise in the
first stage of a preamp. RDH suggests the equivalent input noise resistance
for a triode is about 2.5/gm. Patrick noted he had never seen noise
anywhere near as low as this.

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

Please, please let's all get past the Wikipedia epoch. It's
fine as a starting point, but hopeless as anything else.
It's specifically *not* a reference for *anything*.

Please!

which suggests the 2.5/gm only applies at frequencies above 10KHz and that
below this 1/f noise dominates. It also suggests this can be reduced by
choosing very pure materials for the cathode nickel, and running the tube
at an optimized (generally low) anode current.

The difficulty with valve noise and things that go bump in the
night is that noises were poorly understood in the thirties and
fourties when the text books were written, but the old text books
get quoted now well into the Wikipedia plague, er, epoch.

Eastman, for example, gets it quite wrong, and even Terman
(third edition of Radio Engineering, -the true Bible, not
the imposter bible RDH4- ) is muddled. RDH4 is hopelessly
wrong.

For openers, "shot noise" was defined as BOTH the thermal noise
equivalent of the reciprocal of transconductance AND as a
quantization noise under cathode temperature limited emission,
in the most remotely likely of ancient texts. Could be defined
either way, but not both.

Folks back then didn't have much of a framework for noise
analysis, so it's not too surprising that names might be
misapplied. There's much less excuse for us today to
continue to do the same.


Don't know who the author is and as a rule I take anything on wikipedia with
a large grain of salt. Comments?

There's never enough salt. Arf.





A good framework to think about noise is to cook everything
down into resistances at their working temperatures, and to
calculate the noise from those resistances. This is "thermal"
(or historically Johnson or Nyquist) noise.

Ferzample, a valve's tranconductance engine has a resistance
of reciprocal transconductance at a temperature factor of
2.5 times room temperature for oxide-coated cathodes. This
acts just like a real resistor at that temperature.

Then:

Everything else is a special case, and is largely uncalculatable.
All these "excess" noises arise from material imperfections
and manufacturing imperfections and the other imperfections of
the real world. You'll need to measure them to know anything
believable.

Thanks, as always,

Chris Hornbeck

[Ian Bell]

Chris Hornbeck wrote:



A good framework to think about noise is to cook everything
down into resistances at their working temperatures, and to
calculate the noise from those resistances. This is "thermal"
(or historically Johnson or Nyquist) noise.

Ferzample, a valve's tranconductance engine has a resistance
of reciprocal transconductance at a temperature factor of
2.5 times room temperature for oxide-coated cathodes. This
acts just like a real resistor at that temperature.


Seems reasonable.

Then:

Everything else is a special case, and is largely uncalculatable.
All these "excess" noises arise from material imperfections
and manufacturing imperfections and the other imperfections of
the real world. You'll need to measure them to know anything
believable.


The big question though seems to be just what proportion of the overall
noise are represented by these special cases? My feeling from what Patrick
says he has measured is that they are at least of the same order.

Ian

[Ian Bell]

Phil Allison wrote:


"Ian Bell"

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

which suggests the 2.5/gm only applies at frequencies above 10KHz and
that below this 1/f noise dominates.


** Not what it says at all.

" For special low noise audio tubes, the frequency at which 1/f noise
takes over is reduced as far as possible, maybe to something like a kHz. "

Applies to valves like the 12AX7 and 12AY7 - their noise spectrum is
pretty much white noise, where the energy is *proportional* to bandwidth.
A gradual rise in noise below 1kHz is insignificant, since it is so small
in the first place.


What about the bit about running at low plate currents? Is that correct?


Ian

[Eeyore]

Ian Bell wrote:

Chris Hornbeck wrote:

A good framework to think about noise is to cook everything
down into resistances at their working temperatures, and to
calculate the noise from those resistances. This is "thermal"
(or historically Johnson or Nyquist) noise.

Ferzample, a valve's tranconductance engine has a resistance
of reciprocal transconductance at a temperature factor of
2.5 times room temperature for oxide-coated cathodes. This
acts just like a real resistor at that temperature.


Seems reasonable.

Yes it does.

In semiconductors for example the actual bulk resistance of the silicon must be
taken into account in noise calculations, notably Rbb, the intrinsic base
resistance of the device.

Since the current flow in a valve has to physically flow through the oxide
coating on the cathode, which presumably has some resistance, then that
resistance must surely be incorporated into the noise calculations. Furthermore
of course it's at a high temperature which will worsen the noise.

Graham

[Patrick Turner]

Ian Bell wrote:

A while ago we were discussiig theoretical and actual measured noise in the
first stage of a preamp. RDH suggests the equivalent input noise resistance
for a triode is about 2.5/gm. Patrick noted he had never seen noise
anywhere near as low as this.

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

which suggests the 2.5/gm only applies at frequencies above 10KHz and that
below this 1/f noise dominates. It also suggests this can be reduced by
choosing very pure materials for the cathode nickel, and running the tube
at an optimized (generally low) anode current.

Don't know who the author is and as a rule I take anything on wikipedia with
a large grain of salt. Comments?

Cheers

ian

The reason the noise I see and measure isn't all shot noise
and must be mainly grid input noise is because when have say 2uV of
input noise on a
12AX7 with gain = 70, some 140uV appears at the anode, and this is
easily amplified
by 1,000 to 140mV and viewable on the CRO.

Shot noise is in there, but I don't know how much is is for an average
12AX7,
Ea = 120V, Ia = 0.6mA.

Noise varies from tube to tube and in bandwidth, perhaps cathodes behave
differently...
Gassy old tubes are the noisiest.

Patrick Turner.

[Patrick Turner]

Eeyore wrote:

Phil Allison wrote:

"Ian Bell"

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

which suggests the 2.5/gm only applies at frequencies above 10KHz and that
below this 1/f noise dominates.

** Not what it says at all.

" For special low noise audio tubes, the frequency at which 1/f noise takes
over is reduced as far as possible, maybe to something like a kHz. "

Applies to valves like the 12AX7 and 12AY7 - their noise spectrum is
pretty much white noise, where the energy is *proportional* to bandwidth. A
gradual rise in noise below 1kHz is insignificant, since it is so small in
the first place.

BTW:

The section about " Class AB and B " is a load of ********.

The author say he got his info from " The Audio Amateur" and "
Stereophile" magazine.

Bout as reliable as the weather.....

Hmmmm, the author reckon class AB and B amps produce 'aharmonic distortion' !
There's no explanation either of what the terms mean !

"aharmonic" means not harmonically related. And can be IMD, where the
sum and difference
F produced each side of a centre F are not harmonics of the centre F.
When will readers of Sterophile ever understand IMD? or sidebands, or
envelope modulations?
or any real techno talk?

I don't buy audio mags because they are so utterly full of BS than makes
me want
to throw a brick in the author's direction.

Patrick Turner.


Talk about the blind leading the blind !

Graham

[Patrick Turner]

Ian Bell wrote:

Chris Hornbeck wrote:


A good framework to think about noise is to cook everything
down into resistances at their working temperatures, and to
calculate the noise from those resistances. This is "thermal"
(or historically Johnson or Nyquist) noise.

Ferzample, a valve's tranconductance engine has a resistance
of reciprocal transconductance at a temperature factor of
2.5 times room temperature for oxide-coated cathodes. This
acts just like a real resistor at that temperature.


Seems reasonable.

Then:

Everything else is a special case, and is largely uncalculatable.
All these "excess" noises arise from material imperfections
and manufacturing imperfections and the other imperfections of
the real world. You'll need to measure them to know anything
believable.


The big question though seems to be just what proportion of the overall
noise are represented by these special cases? My feeling from what Patrick
says he has measured is that they are at least of the same order.

Ian

At the end of the day, regardless of how much you either do understand
or don't
understand about noise, you have to minimize it in the circuits you
build.

So try to build something with low noise.

Try to measure it.

Then you know where noise is.

Patrick Turner.

[Eeyore]

Patrick Turner wrote:

Eeyore wrote:

Phil Allison wrote:

"Ian Bell"

I have just read this article:

http://en.wikipedia.org/wiki/Valve_a...er_-_technical

which suggests the 2.5/gm only applies at frequencies above 10KHz and that
below this 1/f noise dominates.

** Not what it says at all.

" For special low noise audio tubes, the frequency at which 1/f noise takes
over is reduced as far as possible, maybe to something like a kHz. "

Applies to valves like the 12AX7 and 12AY7 - their noise spectrum is
pretty much white noise, where the energy is *proportional* to bandwidth. A
gradual rise in noise below 1kHz is insignificant, since it is so small in
the first place.

BTW:

The section about " Class AB and B " is a load of ********.

The author say he got his info from " The Audio Amateur" and "
Stereophile" magazine.

Bout as reliable as the weather.....

Hmmmm, the author reckon class AB and B amps produce 'aharmonic distortion' !
There's no explanation either of what the terms mean !

"aharmonic" means not harmonically related. And can be IMD, where the
sum and difference
F produced each side of a centre F are not harmonics of the centre F.

That's not what he said though !


When will readers of Sterophile ever understand IMD? or sidebands, or
envelope modulations? or any real techno talk?

I don't buy audio mags because they are so utterly full of BS than makes
me want to throw a brick in the author's direction.

Yup.

Graham

[Chris Hornbeck]

On Fri, 22 Jun 2007 07:14:05 GMT, Eeyore
wrote:

Since the current flow in a valve has to physically flow through the oxide
coating on the cathode, which presumably has some resistance, then that
resistance must surely be incorporated into the noise calculations. Furthermore
of course it's at a high temperature which will worsen the noise.

That's a very interesting thought. I always just use a magical
transconductance engine overlay that removes all thought about
emission and space charge and yada-yada, but I'd be interested
in a deeper level if I could find one I could believe.

It's easy to assume that the physics of vacuum valves must
be easier to understand than the quantum physics of semiconductors,
but even that may not be true. I certainly don't know.

Any other thoughts?
Thanks, as always,

Chris Hornbeck

[Patrick Turner]

Bret Ludwig wrote:

When will readers of Sterophile ever understand IMD? or sidebands, or
envelope modulations? or any real techno talk?

I don't buy audio mags because they are so utterly full of BS than makes
me want to throw a brick in the author's direction.


I find that old issues of electronics, rather than audio magazines
(except Audio, nee Audio Engineering) quite useful. Radio and
Television News in the 50s was quite good. ebay has them: I'd like to
scan and put up whole years a la Pete Millett someday.

If you ever get the time.

And who wants to watch a 1950 TV set?

And isn't it better to distil from the past the spirit of the age,
and focus on the best of it, rather than all the many medicocre ways of
doing things?

A 5th Edition of the Radiotron Designer's Handbook would be a real
challenge.

I doubt its really necessary though because anyone who understands
vacuum tubes and analog
systems will design what they like anyway...

In 20 years anything analog will be utterly passe.

Patrick Turner.